WO2023151545A1

WO2023151545A1 - Data storage system, data storage method and apparatus, and related device

Info

Publication number: WO2023151545A1
Application number: PCT/CN2023/074690
Authority: WO
Inventors: 李楚; 叶茂; 冯永刚; 王�锋
Original assignee: 华为技术有限公司
Priority date: 2022-02-08
Filing date: 2023-02-07
Publication date: 2023-08-17
Also published as: CN116610598A

Abstract

Provided is a data storage system, comprising a first device and a second device, the first device comprising a first cache area, and the second device comprising a second cache area, wherein the first device is used for receiving a processing request, writing target data in the data processing request into the second cache area on the basis of RDMA, and writing the target data into the first cache area, wherein after the target data is successfully written into the first cache area and the second cache area, the data processing request is completed in the data storage system; and the second device is used for persistent storage of the target data written into the second cache area. In this way, the response time delay of storing the target data by the data storage system is relatively low, and the target data can be prevented from being lost by means of double-cache, such that the data storage reliability of the data storage system can be improved. In addition, the present application further discloses a corresponding data storage method and apparatus, and a related device.

Description

A data storage system, data storage method, device and related equipment

technical field

The present application relates to the technical field of data storage, and in particular to a data storage system, data storage method, device and related equipment.

Background technique

A data storage system refers to a system that organizes and allocates storage space for storage devices and is responsible for storing and retrieving data. In some practical application scenarios, the data storage system can provide data read and write services in a multi-device mode. Specifically, one of the devices in the data storage system may provide a data access interface externally, and receive new data to be stored through the data access interface. Normally, after the device caches the new data locally, it feeds back an acknowledgment message (acknowledgment, ack) that the data is written successfully to the data sender, and then writes the new data in the local cache to other devices in the data storage system Perform persistent storage to reduce the response delay of the storage data system to store data.

However, when the device that provides the data access interface fails or the cache of the device fails, the data stored in the local cache is lost because it is not persistently stored in time, which reduces the reliability of the data storage system.

Contents of the invention

A data storage system, a data storage method, a device, a computing device, a storage medium and a computer program product are provided, so that the data storage system has relatively low response time delay and high reliability of data storage.

In a first aspect, an embodiment of the present invention provides a data storage system, where the data storage system includes a first device and a second device, and the first device includes a first cache area, and the second device includes a second cache area. Wherein, the first device is used to receive a processing request, and the data processing request may be used to request writing target data, etc., and may be sent to the first device by other devices or an application program on the first device, and the first The device can write the target data in the data processing request into the second cache area based on RDMA, and write the target data into the local first cache area; the second device is used to persist the target data written in the second cache area storage.

After the first device writes the target data into the local first cache area and writes the target data into the second cache area of the second device through RDMA technology, the data processing request is completed. Since RDMA does not require the participation of the CPU, and this solution does not require the second device to persistently store the target data before considering the data processing request to be processed, this makes the response delay of the data storage system to store the target data relatively low. is close to storing data only in the cache area of the first device. At the same time, the target data is stored in the cache areas of the first device and the second device respectively, so that when the cache of one device fails, the data storage system can also avoid the loss of the target data through the cache of the other device, so that This can improve the reliability of data stored in the data storage system. Therefore, the first aspect provides a data storage system with low latency and improved data reliability.

In a possible implementation manner, the first device is further configured to persistently store the target data in the second cache area on the second device after the target data is written into the first cache area and the second cache area Before, generate a response message that the data processing request was successful. In this way, after the target data is successfully written into the two cache areas, the storage system regards the execution of the data processing request as completed, and can generate a response message to the data processing request, so that the response message can be used to send data to According to the feedback from the application processing the request or other devices that the target data has been stored, the response delay for the data storage system to store the target data is low, and the data storage system can persist the target data after feeding back the response message storage.

In a possible implementation manner, the first device is further configured to, after determining that the target data is persistently stored, eliminate the target data in the first cache area. In this way, the storage space occupied by the target data in the first cache area can be released under the condition that the target data is guaranteed to be persistently stored, so that the first device can use the released storage space to store other newly written data later. , realizing recycling of storage resources in the first cache area.

In a possible implementation manner, the second device is further configured to, after determining that the target data is stored persistently, eliminate the target data in the second cache area. In this way, the storage space occupied by the target data in the second cache area can be released, so that subsequent data of the first device can be cached in the released storage space, and the storage resources of the second cache area can be recycled.

In a possible implementation manner, when writing the target data into the second cache area, the first device is specifically configured to generate a WAL including the target data, and then write the WAL into the second cache area based on the RDMA technology. In this way, the second device can persistently store the target data in the WAL by replaying the log.

In a possible implementation manner, the first device is specifically configured to write the WAL including the target data into the first cache area. In this way, when the first device subsequently reads the target data from the local first cache area, the target data may be acquired according to the WAL by playing back a log. Optionally, the first device may also directly write the target data into the first cache area.

In a possible implementation manner, when the first device writes the target data into the first cache area, it is specifically used to generate a key-value pair, where the key in the key-value pair is the identifier of the target data, and the key-value pair The value in is the target data, so that the first device can write the key-value pair into the first cache area. In this way, the first device can generate an index in the first cache area based on the target data, so that when the target data needs to be read, the first device can quickly find the target data from the local first cache area by traversing the index.

In a possible implementation manner, when the target data is written into the first cache area through a key-value pair, the value in the key-value pair also corresponds to the first LSN of the WAL, and the first device is also used to obtain the first LSN of the WAL The second LSN corresponding to the WAL played back by the second device, and when the first LSN is equal to the second LSN, the first device may remove the key-value pair from the first cache area. In this way, the first device can use the LSN corresponding to the WAL to identify which target data in the first cache area is persistently stored, and release the storage space occupied by the target data in the first cache area to support the first device to continue Write other data to realize the recycling of storage resources in the first cache area; and for the data that has not been persisted through the LSN, it can continue to be stored in the first cache area to improve the storage of the data in the data storage system reliability.

In a possible implementation manner, the first device is further configured to receive a data read request, where the data read request includes an identifier of the target data, so that the first device can read the data from the first cache area according to the identifier of the target data Check whether the target data is included, and, when the target data is included in the first cache area, the first device may feed back the found target data. In this way, the first device can quickly find the target data to be read from the local cache area, and can obtain the target data without remotely accessing the second device, thereby effectively improving the efficiency of the data storage system for feeding back target data.

In a possible implementation manner, the first device is further configured to: when the target data is not included in the first cache area, if the target data is eliminated from the first cache area because the persistent storage has been completed, the first device The target data is requested from the second device, so that the first device subsequently feeds back the target data.

In a possible implementation manner, the first device is specifically configured to write the target data into the second cache area based on the unilateral RDMA technology, so that the delay in writing the target data from the first device to the second device can be further effectively reduced , so that the High response efficiency of data storage system to store target data. Optionally, the first device may also write the target data into the second cache area by using a bilateral RDMA technology.

In a possible implementation manner, the data storage system may specifically be a distributed file system, where the first device implements a client in the distributed file system, and the second device implements a server in the distributed file system, And in the distributed file system, the second device persistently stores the target data in a file format.

In a second aspect, an embodiment of the present invention provides a data storage method, the data storage method is applied to a data storage system, the data storage system includes a first device and a second device, and the first device includes a first cache area, the second The second device includes a second cache area, and the method includes: the first device receives a data processing request, and the data processing request includes target data; and, the first device writes the target data into the second cache area based on RDMA, and writes the target data into the second cache area. The first cache area.

In a possible implementation manner, the method further includes: after the target data is written into the first cache area and the target data is written into the second cache area, and after the second Before the device persistently stores the target data in the second cache area, the first device generates a response message indicating that the data processing request is successful.

In a possible implementation manner, the method further includes: after determining that the target data is persistently stored, the first device eliminates the target data in the first cache area.

In a possible implementation manner, the first device writes the target data into the second cache area based on Remote Direct Memory Access (RDMA), including: the first device generates WAL including the target data; the first device writes the WAL into the second cache area based on RDMA. Second cache area.

In a possible implementation manner, the writing the target data into the first cache area by the first device includes: writing the WAL into the first cache area by the first device.

In a possible implementation manner, the first device writes the target data into the first cache area, including: the first device generates a key-value pair, the key in the key-value pair is the identifier of the target data, and the value in the key-value pair is the target data; the first device writes the key-value pair into the first cache area.

In a possible implementation manner, the value in the key-value pair also corresponds to the first log sequence number LSN of the WAL, and the method further includes: the first device obtains the second LSN corresponding to the WAL that has been played back by the second device; When the first LSN is equal to the second LSN, the first device eliminates the key-value pairs in the first cache area.

In a possible implementation manner, the method further includes: the first device receives a data read request, and the data read request includes the identifier of the target data; the first device searches whether the target data is included in the first cache area according to the identifier of the target data data; when the target data is included in the first cache area, the target data is fed back.

In a possible implementation manner, the method further includes: when the first cache area does not include the target data, the first device requests the second device for the target data.

In a possible implementation manner, the first device writing the target data into the second cache area based on remote direct memory access (RDMA) includes: the first device writing the target data into the second cache area based on unilateral RDMA.

In a possible implementation manner, the data storage system includes a distributed file system, the first device implements a client in the distributed file system, the second device implements a server in the distributed file system, and the second device implements the target Data is stored persistently in a file format.

Since the data storage method provided by the second aspect corresponds to the data storage system provided by the first aspect, the technical effects of the second aspect and the implementation methods in the second aspect can be referred to the corresponding first aspect and the first aspect The technical effects of each of the implementation modes are not repeated here.

In the third aspect, the embodiment of the present invention provides a storage device, the storage device is used as the first device in the data storage system, the data storage system further includes a second device, and the storage device includes a first cache area and a processing A device, wherein the first cache area is used to cache data; the processor is used to execute the following method by running a computer program: receiving A data processing request, the data processing request including target data; writing the target data into the second cache area based on remote direct memory access (RDMA), so that the target data in the second cache area is persistently stored; and Writing the target data into the first cache area.

In a possible implementation manner, the processor is further configured to: after the target data is written into the first cache area and the target data is written into the second cache area, and after the Before the target data in the second cache area is persistently stored, a response message indicating that the data processing request is successful is generated.

In a possible implementation manner, the processor is configured to: delete the target data in the first cache area after determining that the target data is persistently stored.

In a possible implementation manner, the processor is configured to: generate a write-ahead log WAL including target data; and write the WAL into the second cache area based on RDMA.

In a possible implementation manner, the processor is configured to: write the WAL into the first cache area; or generate a key-value pair, where a key in the key-value pair is an identifier of the target data , the value in the key-value pair is the target data, and the key-value pair is written into the first cache area.

In a possible implementation manner, the value in the key-value pair also corresponds to the first log sequence number LSN of the WAL, and the processor is further configured to: obtain the WAL corresponding to the playback of the second device the second LSN of the second LSN; when the first LSN is equal to the second LSN, eliminate the key-value pair in the first cache area.

In a possible implementation manner, the processor is further configured to: receive a data read request, where the data read request includes an identifier of the target data; according to the identifier of the target data, search for the first Whether the target data is included in a buffer area; when the target data is included in the first buffer area, the target data is fed back; when the target data is not included in the first buffer area, the target data is sent to the The second device requests the target data.

In a possible implementation manner, the processor is configured to: write the target data into the second cache area based on unilateral RDMA.

In a possible implementation manner, the data storage system includes a distributed file system, the first device implements a client in the distributed file system, and the second device implements a client in the distributed file system The server end, and the second device persistently stores the target data in a file format.

Since the storage device provided by the third aspect corresponds to the first device in the data storage system provided by the first aspect, the third aspect and the technical effects of the implementations in the third aspect can be referred to the corresponding first aspect As well as the technical effects of each implementation manner in the first aspect, details will not be described here.

In a fourth aspect, an embodiment of the present invention provides a computing device, including: a processor and a memory; the memory is used to store instructions, and when the computing device is running, the processor executes the instructions stored in the memory, so that the computing The device executes the second aspect or the data storage method executed by the first device in any implementation manner of the second aspect. It should be noted that the memory may be integrated in the processor, or independent of the processor. A computing device may also include a bus. Wherein, the processor is connected to the memory through the bus. Wherein, the memory may include a readable memory and a random access memory.

In the fifth aspect, the embodiment of the present invention also provides a readable storage medium, the readable storage medium stores a program or an instruction, and when it is run on a computer, any one of the above-mentioned second aspect or the second aspect In the implementation manner, the data storage method executed by the first device is executed.

In a sixth aspect, an embodiment of the present invention further provides a computer program product containing instructions, which, when run on a computer, cause the computer to execute the above-mentioned second aspect or any implementation manner of the second aspect executed by the first device. data storage method.

In addition, for the technical effects brought about by any one of the implementations from the second aspect to the sixth aspect, please refer to the technical effects brought about by different implementations in the first aspect, which will not be repeated here.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some implementations recorded in this application. For example, those skilled in the art can also obtain other drawings based on these drawings.

FIG. 1 is a schematic structural diagram of an exemplary data storage system provided by an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of an exemplary distributed file system provided by an embodiment of the present invention;

FIG. 3 is a schematic flowchart of a data storage method provided by an embodiment of the present invention;

FIG. 4 is a schematic flowchart of a data reading method provided by an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a data storage device provided by an embodiment of the present invention;

Detailed ways

Referring to FIG. 1 , it is a schematic structural diagram of an exemplary data storage system. As shown in FIG. 1, the data storage system 100 may include a device 101 and a device 102, and data interaction may be performed between the device 101 and the device 102, such as data communication through Hypertext Transfer Protocol (Hypertext Transfer Protocol, HTTP).

Wherein, the device 101 can provide a file access interface, such as a portable operating system interface (portable operating system interface, POSIX), etc., so that the device 103 connected to the data storage system 100 or the application (application) 1011 in the device 101 can pass the file The access interface implements operations such as deleting, reading, writing, and modifying files in the data storage 100 . In addition, the device 101 may include a cache area 1, and data may be cached through the cache area 1 .

The device 102 includes a cache area 2 , and the device 102 can cache data through the cache area 2 . Further, the device 102 may also include a persistent storage area, so that the device 102 may write the data in the cache area 2 into the persistent storage area for persistent storage. Exemplarily, the cache area 2 can be implemented by non-volatile memory (non-volatile memory, NVM) such as read-only memory (read-only memory, ROM), flash memory (flash memory), storage class memory (storage class memory, SCM), etc. ), or the cache area 2 may also be implemented by a volatile memory (volatile memory) such as a random-access memory (random-access memory, RAM), which is not limited in this embodiment. The persistent storage area can be realized by a hard disk drive (HDD) or other storage media that can store data for a long time without power supply. The hard disk can be a magnetic disk or a solid-state drive (SSD) Or SCM disk, etc.

When the device 103 or the application 1011 writes new data (hereinafter referred to as target data) to the data storage system 100, if the device 101 only writes the target data into the local cache area 1, although this can reduce the storage capacity of the data storage system 100 The response delay of the target data sent by the application 1011, but when the device 101 fails or the cache area 1 fails, the target data sent by the device 103 or the application 1011 may be lost, which reduces the storage capacity of the data storage system 100. Data reliability.

Based on this, an embodiment of the present invention provides a data storage method, aiming at improving the reliability of data storage while making the data storage system 100 have a smaller response time delay. During specific implementation, for the target data sent by the device 103 or the application 1011, the device 101 not only writes it into the local cache area 1, but also writes the target data into the cache area 2 of the device 102 based on RDMA technology, so that the target The data is stored in the caches of the device 101 and the device 102 at the same time, and the device 102 can then persistently store the target data in the cache area 2 . After the device 101 writes the target data into the local cache area 1 and writes it into the cache area of the device 102 through the RDMA technology, the processing of the data processing request is completed. Since RDMA does not require CPU participation, and the target data can be stored persistently without the device 102 The data processing request is considered to be processed only later, which makes the response delay of the data storage system 100 to store the target data low, and the delay can be close to only storing the data in the cache area 1 of the device 101 . At the same time, the target data is stored in the cache areas of device 101 and device 102 respectively, so that when the target data cache fails on the device 101 side, the data storage system 100 can also prevent the target data from being lost through the cache on the device 102 side. , so that the reliability of data storage in the data storage system 100 can be improved. In this way, while the data storage system 100 has a small response delay, the reliability of data storage is also high.

It should be noted that the system architecture shown in FIG. 1 is only used as an example, and is not intended to limit its specific implementation to this example. For example, in other possible system architectures, the persistent storage area can also be deployed outside the data storage system 100, that is, the device 102 can remotely send the target data in the cache area 2 to the persistent storage area for persistent storage, etc. This embodiment does not limit it.

For example, the data storage system shown in FIG. 1 may specifically be the distributed file system 200 shown in FIG. 2 . As shown in FIG. 2 , the distributed file system 200 includes a client 201 and a server 202 . Wherein, the client 201 may be an application program provided by the distributed file system 200, which may be implemented by the device 101 in FIG. Further, the device 101 where the client is located can also run the application 1011 shown in FIG. 1 , and the application 1011 can create and delete files in the distributed file system 200 through the file access interface provided by the client 201 , read, write, modify and other operations.

The server 202 can be implemented by one or more devices 102 in FIG. 1 , and the one or more devices 102 can be integrated with a memory so that the server 202 can be provided with a cache area 2 through the memory, which can be used for high-speed Cache data; further, the memory integrated on the device 102 can also provide a persistent storage area for the server 202, which can be used for persistent storage of data. Specifically, the device 102 may include multiple storages 1, so that the multiple storages 1 may form a buffer pool; and the device 102 may also include multiple storages 2, so that the multiple storages 2 may build a persistent storage pool. Exemplarily, the memory 1 can be non-volatile memory (non-volatile memory, NVM) such as read-only memory (read-only memory, ROM), flash memory (flash memory), storage class memory (storage class memory, SCM), etc. ), or, the memory 1 may also be a volatile memory (volatile memory) such as a random-access memory (random-access memory, RAM). The memory 2 may be a hard disk drive (hard disk drive, HDD) or a solid-state drive (solid-state drive, SSD) and other devices that can be used for persistent storage of data. In this embodiment, specific implementation manners of the memory 1 and the memory 2 are not limited.

In actual application, the data storage system shown in FIG. 1 may also be applicable to other application scenarios except FIG. 2 , which is not limited in this embodiment. For example, the system structure shown in FIG. 1 may be applicable to an application scenario of centralized storage or an application scenario of distributed storage.

Among them, in the centralized storage application scenario, one or more devices 102 can form a central node, and data is stored centrally in this central node, and all data processing services of the entire data storage system 100 are centrally deployed on this central node superior. At this time, a disk control separation architecture can be adopted between the device 102 and the storage 1 implementing the cache area 2, that is, the device 102 and the storage 1 are deployed independently, or an integrated disk control architecture can be used between the device 102 and the storage 1, that is, the device 102 There may be a slot, and the storage 1 is placed in the device 102 through the slot, and integrated with the device 102 for deployment.

In a distributed storage application scenario, data in the data storage system 100 may be distributed and stored on multiple independent storage nodes. At this point, the device 102 can be integrated with the storage 1 so that the device 102 has computing capability and storage capability at the same time, and a virtual machine can be created on the device 102, or no virtual machine can be created. Alternatively, a storage-computing separation architecture may be adopted between the device 102 and the memory 1 implementing the cache area 2, that is, the device 102 and the memory 1 are deployed independently and communicate through a network. In addition, the memory 1 may include one or more different storage media. In this embodiment This is not limited.

In order to make the above objects, features and advantages of the present invention more comprehensible, various non-limiting implementations in the embodiments of the present invention will be illustrated below in conjunction with the accompanying drawings. Apparently, the described embodiments are some, but not all, embodiments of the present invention. All other embodiments obtained based on the embodiments of the present invention and the above content belong to the protection scope of the present invention.

As shown in FIG. 3 , it is a schematic flowchart of a data storage method in an embodiment of the present invention, and the method can be applied to the data storage system 100 shown in FIG. 1 . In actual application, the method may also be applied to other applicable data storage systems. For ease of understanding and description, the following uses the data storage system 100 shown in FIG. 1 as an example for illustration. The method may specifically include:

S301: The device 101 receives a data processing request, where the data processing request includes target data.

When the application 1011/device 103 writes the target data into the data storage system 100, or modifies the data stored in the data storage system 100 to the target data, it can generate a data processing request including the target data, and pass the data access interface ( Such as POSIX, etc.) sends the data processing request to the device 101 . In this way, the device 101 can analyze the received data processing request to obtain the target data to be stored by the data storage system 100 . In practical applications, the data processing request may also carry operation indication information for the target data, such as write operation, modification operation, etc., so that the device 101 can obtain an operation record for the target data based on the data processing request.

S302: The device 101 writes the target data into the cache area 1 in the device 101 .

S303: The device 101 writes the target data in the data processing request to the cache area 2 in the device 102 based on RDMA.

Wherein, the device 101 may execute step S202 first, and then execute step S203; or, the device 101 may execute step S202 and step S203 at the same time; or, the device 101 may first execute step S203 and then execute step S202, etc. Not limited.

After receiving the target data, the device 101 may write the target data into the local cache area 1 and the cache area 2 in the device 102 . Wherein, when writing the target data into the cache area 2, the device 102 can realize the direct writing of data through remote direct memory access (RDMA) technology, that is, directly write the target data into the cache area 2 through the network , the intervention of the operating system in the device 102 is not required, which not only reduces the impact of data writing on the performance of the operating system of the device 102, but also takes less time to write data into the cache area 2 .

Since it takes less time for the client to write the target data into the cache area 1 and the cache area 2 , the response delay for the data storage system 100 to cache the target data is relatively small. It is worth noting that when the device 101 successfully writes the target data into the cache area 1 and the cache area 2, the device 101 can feed back to the application 1011 or the device 103 that the data processing request has been processed, that is, the storage of the target data is completed. Specifically, the device 101 can generate a response message indicating that the data processing request is successful, and send the response message to the application 1011 or the device 103, so that the data storage system 100 can notify the application 1011 before actually performing persistent storage on the target data. Or the device 102 finishes storing the target data. In this way, the response efficiency of the data storage system 100 to the application 1011 or the device 103 can be effectively improved and the response delay can be reduced. In practical applications, the data storage system 100 may complete the persistent storage of the target data in a future time period. In a further possible implementation, the device 101 may write the target data into the cache area 2 based on a one-sided RDMA technology, so as to further reduce the time delay for the device 101 to write data into the cache area 2, thereby The overall time delay for the data storage system 100 to respond to the data processing request can be further reduced.

Moreover, the target data is stored in the caches of the device 101 and the device 102 at the same time, which makes it possible for the data storage system 100 to use the target data in the cache of the other party to avoid data loss when there is a problem with one of the caches. The reliability of data storage in the data storage system 100 can be improved.

In an actual application scenario, the device 102 may support multiple devices 101 to read and write data, for example, the device 102 may support multiple clients 201 to read and write data. Therefore, the device 102 may pre-determine the corresponding cache areas 2 for multiple different devices 101 , and the cache areas 2 corresponding to different devices 101 do not overlap. In this way, problems such as data overwriting when different devices 101 write data into the cache of the device 102 can be avoided. As an implementation example, before sending target data to device 102, device 101 may pre-apply for cache space from device 102, such as sending a cache allocation request including device 101 identifier to device 102, so as to request device 102 to allocate The corresponding cache area 2 is allocated, and the address space of the cache area 2 is fed back to the device 101 . In this way, the subsequent device 101 can directly write the target data into the cache area 2 in the device 102 based on the address space.

In this embodiment, the following six implementations of caching target data in the device 101 and the device 102 are provided:

In a first possible implementation manner, before writing the target data into the local cache area 1, the device 101 may pre-write the target data and the operation instruction information for the target data into a write-ahead log (write- ahead logging, WAL), so that the device 101 can write the WAL including the target data into the local cache area 1, and directly write the WAL including the target data through two-sided (two-sided) RDMA technology or unilateral RDMA technology into cache area 2 in device 102.

In a second possible implementation manner, device 101 writes WAL including target data into cache area 1, and directly writes the target data into cache area 2 through RMDA (including unilateral RDMA or bilateral RDMA) technology.

In a third possible implementation manner, the device 101 may directly write the target data into the local cache area 1, and write the WAL including the target data into the cache area 2 through the RDMA technology.

In a fourth possible implementation manner, the device 101 directly writes the target data into the cache area 1, and directly writes the target data into the cache area 2 through RMDA technology.

In a fifth possible implementation manner, the device 101 directly writes the target data into the cache area 2 through the RDMA technology, and generates a key-value pair (key-value) according to the target data, wherein the key in the key-value pair (key ) is the identification of the target data, and the value (value) in the key-value pair is the target data, so that the device 101 can use the key-value pair as a data query index and write it into the local cache area 1, so that the subsequent query target data. In practical applications, for multiple data sent by the application 1011, the device 101 can generate corresponding key-value pairs for each data, so that the device 101 can generate and save corresponding indexes based on multiple key-value pairs, such as building a tree Indexes to structures, etc. In this way, when the device 101 saves the key-value pair in the local cache area 1, the device 101 can add the key-value pair to the index to update the index, so that the subsequent device 101 can search for the key-value pair by traversing the index. to the data that the application 1011 needs to read.

In a sixth possible implementation manner, the device 101 writes the WAL including the target data into the cache area 2 through the RDMA technology, and writes the key-value pairs generated according to the target data into the cache area 1 .

It should be noted that the above-mentioned implementation methods of caching the target data in the device 101 and the device 102 are only for some exemplary illustrations. In practical applications, the device 101 can also realize the caching of the target data in the device 101 and the device 102 in other ways. The embodiment does not limit this.

Normally, the device 101 may generate multiple WALs within a period of time, for example, the application 1011 continuously sends multiple data within the period of time, or multiple processes/threads of the application 1011 send multiple data in parallel. During this process, the device 101 may write the multiple WALs into the cache area 2 one by one, that is, each time a WAL is generated, the device 101 writes the WAL into the cache area 2 of the device 102 . Alternatively, the client 102 can also write multiple WALs into the cache area 2 in batches. For example, the device 101 can write 10 WALs in batches to the cache area 2 each time, so as to reduce the number of writes to the cache area 2. The number of communications between the device 101 and the device 102 is the number of WALs. Among them, when writing multiple WALs in batches, the device 101 can periodically send multiple WALs in batches to the cache area 2 of the device 102, for example, the device 101 can write multiple WALs generated sequentially within 1 second into the cache area in batches 2, etc.; or, when multiple processes/threads of the application 1011 send multiple When storing data, the device 101 can generate multiple WALs in parallel, so that the device 101 can write the multiple WALs generated in parallel to the device 102 in batches.

As an implementation example, the device 101 may encapsulate multiple operations performed inside the data storage system 100 into a single semantic, thereby further reducing the number of WALs sent by the device 101 to the device 102 . Specifically, the operation instruction issued by the application 1011 instructs to perform numerical scaling on all data belonging to the user according to the preset scaling ratio. That is, multiple scaling operations are performed. At this point, device 101 can generate a single semantic-level log for the multiple scaling operations performed on the multiple data, and write the semantic-level log into the cache area 2 of device 102, which is different from the multiple scaling operations performed separately In terms of the manner of generating multiple logs, the number of logs generated by the device 101 can be effectively reduced, so that the number of logs sent by the device 101 to the device 102 can be effectively reduced. Correspondingly, when playing back the log, the device 102 can play back multiple operations according to the semantics recorded in the log.

Further, before caching the target data, the device 101 may also determine whether the operation on the target data is a legal operation according to the received data processing request. When it is determined that the operation on the target data is a legal operation, the device 101 executes the operation and stores the target data in the cache area 1 and the cache area 2; and when it is determined that the operation on the target data is an illegal operation, the device 101 It is possible to refuse to perform the operation and refuse to store the target data, etc. For example, when the operation on the target data exceeds the scope of operation authority of the application 1011, the device 101 may determine that the operation is illegal, so that the device 101 refuses to perform the operation and refuse to store the target data. For another example, when the operation on the target data is the operation of creating a file, if the device 101 determines that the name of the file created based on the target data is the same as that of other files, the device 101 may also determine that the operation is illegal and refuse to execute The operation to create a file.

In actual application, after the device 101 writes the target data into the local cache area 1 and the cache area 2 of the device 102, it can return a feedback result of successful operation to the application 1011 to indicate to the application 1011 that the target data has been successfully stored in the data storage system 100. In this way, for the application 1011, the response delay of the data storage system 100 for storing the target data is small, which meets the requirement of the application 1011 for low-latency response.

S304: The device 102 persistently stores the target data in the cache area 2 .

Usually, the address space in the cache area 2 is smaller than the address space used by the device 102 for persistently storing data, therefore, the device 102 can downwrite the data stored in the cache area 2 to the persistent storage area. Wherein, when the data stored in the cache area 2 is WAL including the target data, the device 102 can obtain the target data by playing back the WAL, and write it into the memory 2 .

In a possible implementation manner, the device 102 may periodically scan the cache area 2 corresponding to each device 101 , and when there is data in the cache area 2 , the device 102 may persistently store the data in the memory 2 . Or, when the amount of data stored in the cache area 2 reaches a preset threshold, the device 102 may persistently store the data in the cache area 2 into the memory 2, and when the amount of data stored in the cache area 2 does not reach When the threshold is preset, the device 102 may not persistently store the data in the current scan cycle, and then perform persistent storage of the data when it is subsequently determined that the amount of data in the cache area 2 reaches the preset threshold.

In another possible implementation, since the data in the cache area 2 is written by the device 101, the device 101 can notify the device 102 to update the cache area 2 after writing the target data into the cache area 2. 2 for persistent storage of data. Alternatively, the device 101 may periodically notify the device 102 to persistently store the data in the cache area 2 . Wherein, when the device 101 does not write data into the cache area 2 in a period, the device 101 may not need to send a notification of persistently stored data to the device 102 in this period.

In actual application, the data storage system 100 may also trigger the device 102 to persistently store the data in the cache area 2 in other ways, which is not limited in this embodiment.

It can be understood that in actual application scenarios, the data storage capacity of the cache area 1 and the cache area 2 is limited, and it is difficult to support the device 101 and the device 102 to store an excessive amount of data. Therefore, in a further possible implementation, after the target data is persisted After being stored in the memory 2, the device 101 and/or the device 102 may delete the target data in the cache area 1, so as to release the storage space occupied by the target data in the cache area 1.

As an implementation example, when WAL including the target data is stored in the cache area 1, the device 101 may record a first log serial number (log serial number, LSN) corresponding to the target data when generating the WAL. Wherein, the LSN of the WAL is generally a monotonically increasing positive integer, that is, for multiple WALs generated sequentially, the LSNs corresponding to the multiple WALs increase gradually. After the device 102 persistently stores the target data in the cache area 2 to the memory 2, it may feed back the second LSN corresponding to the played back WAL to the device 101 . In this way, the device 101 can determine, based on the received second LSN, that all WALs with LSNs smaller than the second LSN have completed playback, so that all data included in the WALs with LSNs smaller than the second LSN have been written into the memory 2 . At this time, the device 101 can compare the size between the first LSN and the second LSN, and when the first LSN is equal to the second LSN, it indicates that the WAL including the target data has completed playback, and the target data has been persistently stored in In the memory 2, at this time, the device 101 can eliminate the target data in the cache area, for example, can eliminate the WAL including the target data, etc., and release the storage space occupied by the target data in the cache area 1, so as to support the device 101 to the More other data is stored in cache area 1.

In actual application, data written by the application 1011 multiple times may be stored in the cache area 1, and the device 101 may generate a WAL including the data each time the data written by the application 1011 is stored. In this way, device 101 can eliminate one or more WALs whose LSN is less than or equal to the second LSN, and WALs whose LSN is greater than the second LSN can continue to be stored in cache area 1 without being eliminated.

In another implementation example, when the target data is stored in the cache area 1 in the form of a key-value pair and the WAL including the target data is stored in the cache area 2, the value in the key-value pair may correspond to the first value of the WAL. An LSN, the first LSN may be recorded in the local cache by the device 101 when generating the WAL. After the device 102 persistently stores the target data in the cache area 2 to the memory 2, it may feed back the second LSN corresponding to the played back WAL to the device 101 . In this way, device 101 can compare the size between the first LSN corresponding to the value in the key-value pair and the second LSN corresponding to the WAL that device 102 has played back. When the first LSN is equal to the second LSN, the characterization target data has been persistently stored in the storage 2 by the server, so that the device 101 can eliminate the key-value pair in the storage area 1 to release the key-value pair in the cache area 1 The storage space used in . When the first LSN is greater than the second LSN, it indicates that the target data has not been persistently stored, and at this time, the key-value pair can continue to be stored in the cache area 1 .

It is worth noting that the various implementation examples above are only used as an illustration. In actual application, the device 101 may also determine whether the target data has been persistently stored in other ways, such as adding a gradually increasing Timestamps, serial numbers, etc. determine whether to eliminate target data from the cache area 1, which is not limited in this embodiment.

In addition, the device 102 may also delete the target data in the cache area 2 after the target data is persistently stored, so as to reclaim the storage space occupied by the target data in the cache area 2 . For example, when the target data is stored in the cache area 2 in the form of WAL, after the device 102 completes the playback of the WAL and persistently stores the target data, it can eliminate the WAL whose LSN is smaller than the above-mentioned second LSN, and release the WAL The occupied storage space, in order to use the released storage space to continue to store other WAL newly written by the device 101. Wherein, after the device 102 eliminates the WAL whose LSN is smaller than the above-mentioned second LSN, it can feed back the address information corresponding to the released storage space to the device 101, so that the subsequent device 101 can write the newly generated WAL into the released storage according to the address information. space.

The above-mentioned embodiments are illustrated by taking the storage of target data in the device 101 and the device 102 as an example. Normally, the application 1011 or the device 103 may also read the target data stored in the data storage system 100 . Combine below FIG. 4 , taking the application 1011 reading the target data stored in the data storage system 100 as an example, exemplarily illustrates the data reading process. Referring to FIG. 4, a schematic flow diagram of a data reading method is shown, and the method may specifically include:

S401: The device 101 receives a data read request sent by the application 1011, where the data read request includes an identifier of target data.

S402: The device 101 searches whether the target data is stored in the local cache area 1 according to the identifier of the target data. If found, continue to execute step S306; and if not found, continue to execute step S303.

S403: The device 101 requests the device 102 to feed back the target data.

It can be understood that, during the process of storing target data, the data storage system 100 will first cache the target data in the local cache area 1 of the device 101 . Therefore, the device 101 may first search from the local cache area 1 whether the target data is still stored in the local cache. In this way, if the device 101 finds the target data, the device 101 can directly feed back the target data in the cache area 1 to the application 1011 , thereby effectively reducing the time delay for the data storage system 100 to respond to the application 1011 . In addition, since the data stored in the local cache area 1 of the device 101 has a certain timeliness, that is, the data may be eliminated by the device 101 after being stored in the cache area 1 for a period of time, therefore, when the device 101 does not find the The target data indicates that the target data has been persistently stored by the device 102 , so that the device 101 can request the target data from the device 102 .

As some examples, this embodiment provides the following exemplary implementation manners of searching for target data in cache area 1:

In the first implementation example, the target data is directly stored in the cache area 1, and the device 101 can search the target data from the cache area 1 according to the identifier of the target data by traversing the data in the cache area 1.

In the second implementation example, the target data is stored in the cache area 1 in the form of WAL, then the device 101 can determine whether there is a WAL containing the target data according to the identifier of the target data by traversing each WAL in the cache area 1, Therefore, the device 101 can obtain the target data by playing back the WAL.

In the third implementation example, the target data is stored in the cache area 1 in the form of a key-value pair, then the device 101 can use the identifier of the target data as a key to traverse the pre-built index to find out the A key-value pair, so that the value in the key-value pair is the target data to be searched.

In actual application, the device 101 may also search for target data from the local cache area 1 in other ways, which is not limited in this embodiment.

S404: The device 102 searches for the target data from the persistent storage area.

Wherein, the specific process of device 102 searching for target data from the persistent storage area may refer to the implementation manner of device 101 searching for target data in cache area 1, which will not be described in detail here in this embodiment.

S405: The device 102 feeds back the found target data to the device 101.

S406: The device 101 feeds back the target data to the application 1011.

The data storage method and the data reading method provided by the present invention are described in detail above with reference to FIG. 1 to FIG. 4 , and the computing device provided according to the present invention will be described below in conjunction with FIG. 5 .

With the same inventive concept as the above method, the embodiment of the present invention also provides a computing device. As shown in FIG. 5 , the computing device 500 may include a communication interface 510 and a processor 520 . Optionally, the computing device 500 may further include a memory 530 . Wherein, the memory 530 may be set inside the computing device 500 , and may also be set outside the computing device 500 . Exemplarily, each action performed by the first device in the embodiments shown in FIG. 3 and FIG. 4 may be implemented by the processor 520 . The processor 520 may obtain the data processing request through the communication interface 510, and use it to implement any method executed in FIG. 3 and FIG. 4 . In the implementation process, each step of the processing flow can be implemented through the integrated logic circuit of the hardware in the processor 520 Or instructions in the form of software complete the methods executed in FIG. 3 and FIG. 4 . For the sake of brevity, details are not repeated here. The program codes executed by the processor 520 to implement the above methods may be stored in the memory 530 . The memory 530 is connected to the processor 520, such as a coupled connection.

Some features of the embodiments of the present invention may be implemented/supported by the processor 520 executing program instructions or software codes in the memory 530 . The software components being loaded on the memory 530 can be summarized functionally or logically, for example, the data writing module 502, the eliminating module 503, and the searching module 504 shown in FIG. 5 . The function of the communication module 501 can be realized by the communication interface 510 .

Any communication interface involved in the embodiments of the present invention may be a circuit, a bus, a transceiver or any other device that can be used for information exchange. For example, the communication interface 510 in the computing device 500 , for example, the other device may be a device connected to the computing device 500 .

The processor involved in the embodiment of the present invention may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or Execute the methods, steps and logic block diagrams disclosed in the embodiments of the present invention. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the methods disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor.

The coupling in the embodiments of the present invention is an indirect coupling or communication connection between devices, modules or modules, which may be in electrical, mechanical or other forms, and is used for information exchange between devices, modules or modules.

The processor may operate in conjunction with the memory. The memory may be a nonvolatile memory, such as a hard disk or a solid state disk, or a volatile memory, such as a random access memory. A memory is, but is not limited to, any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.

The embodiment of the present invention does not limit the specific connection medium among the communication interface, the processor, and the memory. For example, the memory, the processor, and the communication interface can be connected through a bus. The bus can be divided into address bus, data bus, control bus and so on.

Based on the above embodiments, an embodiment of the present invention also provides a storage device. The storage device is used as the first device in the data storage system, and the data storage system further includes a second device. Exemplarily, the data storage system can be, for example, It is the system shown in FIG. 1 or FIG. 2 or the like. Wherein, the storage device includes a first cache area (such as the cache area 1 in the above embodiment) and a processor, the first cache area is used for caching data, and the processor is used to execute the following method by running a computer program:

receiving a data processing request, the data processing request including target data;

Writing the target data into the second cache area based on remote direct memory access RDMA, so that the target data in the second cache area is persistently stored; and writing the target data into the first cache area .

Further, the processor executes the method performed by the device 101 provided in any one or more of the foregoing embodiments through a computer program.

Based on the above embodiments, an embodiment of the present invention also provides a computer storage medium, in which a software program is stored, and when the software program is read and executed by one or more processors, any one or more of the above-mentioned The method executed by the data storage system 100 provided in the embodiment. The computer storage medium may include: various media capable of storing program codes such as U disk, mobile hard disk, read-only memory, random access memory, magnetic disk or optical disk.

Based on the above embodiments, an embodiment of the present invention also provides a chip, which includes a processor, configured to implement the functions of the data storage system 100 involved in the above embodiments, for example, to implement the functions executed in FIG. 3 and FIG. 4 Methods. Optionally, the chip further includes a memory for necessary program instructions and data executed by the processor. The chip may consist of chips, or may include chips and other discrete devices.

Those skilled in the art should understand that the embodiments of the present invention may be provided as methods, systems, or computer program products. Accordingly, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

The terms "first", "second" and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, and this is merely a description of the manner in which objects with the same attribute are described in the embodiments of the present application.

Apparently, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the scope of the embodiments of the present invention. In this way, if the modifications and variations of the embodiments of the present invention fall within the scope of the claims of the application and their equivalent technologies, the application also intends to include these modifications and variations.

Claims

A data storage system, characterized in that the data storage system includes a first device and a second device, the first device includes a first cache area, and the second device includes a second cache area;

The first device is configured to receive a data processing request, and write target data in the data processing request into the second cache area based on Remote Direct Memory Access (RDMA), and write the target data into the first cache area. cache area;

The second device is configured to persistently store the target data in the second cache area.
The data storage system according to claim 1, wherein the first device is further used for:

After the target data is written into the first cache area and the target data is written into the second cache area, and the second device persists the target data in the second cache area Before storage, generate a response message indicating that the data processing request is successful.
The data storage system according to claim 1 or 2, wherein the first device is further used for:

After the target data is persistently stored, the target data in the first cache area is eliminated.
The data storage system according to any one of claims 1 to 3, wherein the first device is specifically used for:

Generate a write-ahead log WAL that includes the target data;

Writing the WAL into the second cache area based on RDMA.
The data storage system according to claim 4, wherein the first device is specifically used for:

writing the WAL into the first cache area; or,

generating a key-value pair, where the key in the key-value pair is the identifier of the target data, and the value in the key-value pair is the target data; and writing the key-value pair into the first cache area.
The data storage system according to claim 5, wherein the value in the key-value pair also corresponds to the first log sequence number LSN of the WAL, and the first device is also used for:

Obtain a second LSN corresponding to the WAL that has been played back by the second device;

When the first LSN is equal to the second LSN, the key-value pair in the first cache area is eliminated.
The data storage system according to any one of claims 1 to 6, wherein the first device is further used for:

receiving a data read request, where the data read request includes an identifier of the target data;

According to the identification of the target data, first search whether the target data is included in the first cache area;

when the target data is included in the first cache area, feeding back the target data;

When the target data is not included in the first cache area, request the target data from the second device.
The data storage system according to any one of claims 1 to 7, wherein the first device is specifically configured to write the target data into the second cache area based on unilateral RDMA.
The data storage system according to any one of claims 1 to 8, wherein the data storage system includes a distributed file system, the first device implements a client in the distributed file system, and the The second device implements the server in the distributed file system, and the second device persistently stores the target data in a file format.
A data storage method, characterized in that the data storage method is applied to a data storage system, the data storage system includes a first device and a second device, the first device includes a first cache area, and the second The device includes a second cache area, the method includes:

The first device receives a data processing request, the data processing request including target data;

The first device writes the target data into the second cache area based on Remote Direct Memory Access (RDMA);

writing the target data into the first cache area by the first device;

The second device persistently stores the target data in the second cache area.
The method according to claim 10, characterized in that the method further comprises:

After the target data is written into the first cache area and the target data is written into the second cache area, and the second device persists the target data in the second cache area Before storage, the first device generates a response message indicating that the data processing request is successful.
The method according to claim 10 or 11, wherein the method further comprises:

After the target data is persistently stored, the first device eliminates the target data in the first cache area.
The method according to any one of claims 10 to 12, wherein the first device writes the target data into the second cache area based on Remote Direct Memory Access (RDMA), comprising:

said first device generates a write-ahead log WAL comprising target data;

The first device writes the WAL into the second cache area based on RDMA.
The method according to claim 13, wherein writing the target data into the first cache area by the first device comprises:

The first device writes the WAL into the first cache area; or,

The first device generates a key-value pair, the key in the key-value pair is the identifier of the target data, and the value in the key-value pair is the target data; write the key-value pair into the The first cache area.
A storage device, characterized in that the storage device is used as a first device in a data storage system, and the data storage system further includes a second device, and the storage device includes:

The first cache area is used to cache data;

Processor for performing the following methods by running a computer program:

receiving a data processing request, the data processing request including target data;

Writing the target data into the second cache area based on remote direct memory access RDMA, so that the target data in the second cache area is persistently stored; and writing the target data into the first cache area .
The storage device according to claim 15, wherein the processor is further configured to:

After the target data is written into the first cache area and the target data is written into the second cache area, and before the target data in the second cache area is persistently stored, generating the A successful response message to the above data processing request.
The storage device according to claim 15 or 16, wherein the processor is configured to:

After it is determined that the target data is persistently stored, the target data in the first cache area is eliminated.
The storage device according to any one of claims 15 to 17, wherein the processor is configured to:

Generate a write-ahead log WAL that includes the target data;

Writing the WAL into the second cache area based on RDMA.
The storage device according to claim 18, wherein the processor is configured to:

writing the WAL into the first cache area; or,

generating a key-value pair, where the key in the key-value pair is the identifier of the target data, and the value in the key-value pair is the target data; and writing the key-value pair into the first cache area.
A computing device, characterized in that the computing device includes a processor and a memory;

The processor is configured to execute instructions stored in the memory, so that the device executes the method performed by the first device according to any one of claims 10 to 14.
A computer-readable storage medium, characterized by comprising instructions, the instructions are used to implement the method performed by the first device according to any one of claims 10 to 14.